Google unveils Deep Research Max tier

Google introduced Deep Research Max on April 21 as part of a broader AI product push tied to Cloud Next ’26, adding a higher-end autonomous research tier to its Gemini lineup. The company said the new offering sits alongside Deep Research and is aimed at long-running investigations that pull from public web sources and private workspace data. Google also grouped the release with Gemini Enterprise Agent Platform, Gemma 4 and eighth-generation TPUs in its official April AI recap. ### When did Google actually announce Deep Research Max? Google’s Deep Research Max announcement appeared on April 21 in a post on The Keyword, not on May 19. In that post, Google described Deep Research and Deep Research Max as “two new evolutions” of its autonomous research agent and said both were built around Gemini 3.1 Pro. Google’s April 2026 recap, published later, folded Deep Research Max into the company’s larger month of AI launches. (blog.google) That recap listed Deep Research Max, Gemma 4, Gemini Enterprise Agent Platform and eighth-generation TPUs together as part of the same April release cycle. ### What is Deep Research Max supposed to do? Deep Research Max is presented by Google as the top-end version of its autonomous research agent. (blog.google) Google’s developer documentation says it is “optimized for long-running, accuracy-critical investigations” that synthesize information from “hundreds of public web sources and private workspace data” into cited reports. Google’s product materials say the tier supports collaborative planning, visualization, MCP servers and File Search. (blog.google) In a separate blog post, the company said the broader Deep Research product line had moved from summarization toward enterprise workflows in areas including finance, life sciences and market research. ### Which Gemini model and features are attached to it? Gemini 3.1 Pro is the model Google tied directly to Deep Research Max. (ai.google.dev) Google said the integration of Gemini 3.1 Pro was central to the new Deep Research and Deep Research Max releases. Google’s materials also point to multimodal and visualization features around the product. The developer page says Deep Research Max supports visualization, while Google’s broader Gemini and Cloud posts describe Agent Platform access to Gemini 3.1 Pro and other multimodal model options. (ai.google.dev) The social framing that connected Deep Research Max to “multimodal features in Gemini 3.1” is consistent with those official product descriptions, though Google’s own Deep Research Max page describes the research workflow more specifically in terms of cited synthesis, File Search and MCP integrations. (blog.google) ### How does this fit with Gemini Enterprise Agent Platform? Gemini Enterprise Agent Platform was announced by Google Cloud in April as a system to build, scale, govern and optimize agents. Google said the platform combines Vertex AI’s model and agent-building tools with new features for integration, orchestration, DevOps and security. Cloud Next ’26 materials place Agent Platform and eighth-generation TPUs at the center of Google’s “agentic” enterprise pitch. (ai.google.dev) Google’s April recap then positioned Deep Research Max within that same package of launches, suggesting the company sees the research tier as one piece of a larger enterprise-agent stack. That linkage is an inference from how Google grouped the products in its official recaps. (cloud.google.com) ### Where does Gemma 4 enter the picture? Gemma 4 was released earlier in April as Google’s latest open model family. Google described it as its “most capable open model family yet” and later included it in the same April AI roundup that mentioned Deep Research Max. Google’s official posts do not say Deep Research Max is built on Gemma 4. Instead, the direct model linkage for Deep Research Max is Gemini 3.1 Pro, while Gemma 4 appears in the same launch window as a separate part of Google’s AI portfolio. (blog.google) ### What is the next concrete place to look? Google’s developer documentation for the Deep Research Max preview is the clearest source for current product details, including supported features such as MCP servers and File Search. (blog.google) Google’s Cloud Next ’26 recap and Gemini Enterprise Agent Platform posts also remain the main official references for how the company is packaging the broader agent push. (ai.google.dev)